Suppose I am solving the following limit:
$$\lim_{(x,y)\rightarrow(1, -1)} \frac{x^2 - y^2}{x^3 + y^3}$$
I begin by factorizing as follows:
$$\lim_{(x,y)\rightarrow(1,-1)} \frac{(x+y)(x-y)}{(x+y)(x^2-xy+y^2)}$$
And then I am tempted to factor out $(x+y)$ from the numerator and denominator:
$$\lim_{(x,y)\rightarrow(1,-1)} \frac{(x-y)}{(x^2-xy+y^2)}$$
Normally when factoring out the terms we must check that they do not equal zero. Sometimes in the context of limits though, we can disregard this concern if the term being factored out is only approaching zero without ever reaching it. However in the example above it seems that we must still consider this, since if we approach $(1,-1)$ along the line $x = -y$, then the term $(x + y)$ does indeed equal 0, even before it reaches the origin.
What is going on here?
1) When evaluating the limit in $\mathbb{R}^2$, do we only care about approach paths which are actually defined as the approach the point in question? (the path $(x = -y)$ in this case is undefined at all points!)
2) Does the existence of this approach path for which a limit doesn't exist (because nothing on that path is defined anyway!) mean that the limit itself at that point doesn't exist?